Chip War: The Fight for the World's Most Critical Technology

by Chris Miller

If Beijing succeeds, it will remake the global economy and reset the balance of military power. World War II was decided by steel and aluminum, and followed shortly thereafter by the Cold War, which was defined by atomic weapons. The rivalry between the United States and China may well be determined by computing power. Strategists in Beijing and Washington now realize that all advanced tech—from machine learning to missile systems, from automated vehicles to armed drones—requires cutting-edge chips, known more formally as semiconductors or integrated circuits. A tiny number of companies control their production.

typical chip might be designed with blueprints from the Japanese-owned, UK-based company called Arm, by a team of engineers in California and Israel, using design software from the United States. When a design is complete, it’s sent to a facility in Taiwan, which buys ultra-pure silicon wafers and specialized gases from Japan. The design is carved into silicon using some of the world’s most precise machinery, which can etch, deposit, and measure layers of materials a few atoms thick. These tools are produced primarily by five companies, one Dutch, one Japanese, and three Californian, without which advanced chips are basically impossible to make. Then the chip is packaged and tested, often in Southeast Asia, before being sent to China for assembly into a phone or computer.

Chips from Taiwan provide 37 percent of the world’s new computing power each year. Two Korean companies produce 44 percent of the world’s memory chips. The Dutch company ASML builds 100 percent of the world’s extreme ultraviolet lithography machines, without which cutting-edge chips are simply impossible to make. OPEC’s 40 percent share of world oil production looks unimpressive by comparison.

Asia’s vast pool of cheap labor attracted chipmakers looking for low-cost factory workers.

Washington’s foreign policy strategists embraced complex semiconductor supply chains as a tool to bind Asia to an American-led world.

Because vacuum tubes glowed like lightbulbs, they attracted insects, requiring regular “debugging” by their engineers.

Kilby called his invention an “integrated circuit,” but it became known colloquially as a “chip,” because each integrated circuit was made from a piece of silicon “chipped” off a circular silicon wafer.

The eight defectors from Shockley’s lab are widely credited with founding Silicon Valley. One of the eight, Eugene Kleiner, would go on to found Kleiner Perkins, one of the world’s most powerful venture capital firms.

Gordon Moore, who went on to run Fairchild’s R&D process, would later coin the concept of Moore’s Law to describe the exponential growth in computing power.

The challenge would be making chips that civilians could afford. The military paid top dollar, but consumers were price sensitive. What remained tantalizing, though, was that the civilian market was far larger than even the bloated budgets of the Cold War Pentagon. “Selling R&D to the government was like taking your venture capital and putting it into a savings account,” Noyce declared. “Venturing is venturing; you want to take the risk.”

Fairchild, however, was still owned by an East Coast multimillionaire who paid his employees well but refused to give them stock options, viewing the idea of giving away equity as a form of “creeping socialism.” Eventually, even Noyce, one of Fairchild’s cofounders, began wondering whether he had a future at the firm. Soon everyone began looking for the exit.

Yet the two countries’ semiconductor systems couldn’t have been more different. Whereas Silicon Valley’s startup founders job-hopped and gained practical “on the factory floor” experience, Shokin called the shots from his ministerial desk in Moscow. Yuri Osokin, meanwhile, lived in obscurity in Riga, highly respected by his colleagues but unable to speak about his invention with anyone who lacked a security clearance. Young Soviet students didn’t pursue electrical engineering degrees, wanting to be like Osokin, because no one knew that he existed. Career advancement required becoming a better bureaucrat, not devising new products or identifying new markets.

Washington adopted an official policy that “a strong Japan is a better risk than a weak Japan.” Apart from a short-lived effort to shut down Japan’s research into nuclear physics, the U.S. government supported Japan’s rebirth as a technological and scientific power. The challenge was to help Japan rebuild its economy while binding it to an American-led system. Making Japan a transistor salesman was core to America’s Cold War strategy.

Chip firms hired women because they could be paid lower wages and were less likely than men to demand better working conditions. Production managers also believed women’s smaller hands made them better at assembling and testing finished semiconductors.

From South Korea to Taiwan, Singapore to the Philippines a map of semiconductor assembly facilities looked much like a map of American military bases across Asia.

Noyce and Moore abandoned Fairchild as quickly as they’d left Shockley’s startup a decade earlier, and founded Intel, which stood for Integrated Electronics.

He proposed coupling a tiny transistor with a capacitor, a miniature storage device that is either charged (1) or not (0). Capacitors leak over time, so Dennard envisioned repeatedly charging the capacitor via the transistor. The chip would be called a dynamic (due to the repeated charging) random access memory, or DRAM. These chips form the core of computer memory up to the present day.

By contrast, the other main type of chips—those tasked with “computing” rather than “remembering”—were specially designed for each device, because every computing problem was different. A calculator worked differently than a missile’s guidance computer, for example, so until the 1970s, they used different types of logic chips. This specialization drove up cost, so Intel decided to focus on memory chips, where mass production would produce economies of scale.

The Japanese “pay 6 percent, maybe 7 percent, for capital. I pay 18 percent on a good day,” he complained. Building advanced manufacturing facilities was brutally expensive, so the cost of credit was hugely important. A next-generation chip emerged roughly once every two years, requiring new facilities and new machinery. In the 1980s, U.S. interest rates reached 21.5 percent as the Federal Reserve sought to fight inflation. By contrast, Japanese DRAM firms got access to far cheaper capital. Chipmakers like Hitachi and Mitsubishi were part of vast conglomerates with close links to banks that provided large, long-term loans. Even when Japanese companies were unprofitable, their banks kept them afloat by extending credit long after American lenders would have driven them to bankruptcy.

Meanwhile, tight restrictions on stock markets and other investments left people with little choice but to stuff savings in bank accounts. As a result, banks were flush with deposits, extending loans at low rates because they had so much cash on hand.

But after the two countries had signed a mutual defense pact in 1951, the U.S. began cautiously to encourage Japanese rearmament, seeking military support against the Soviet Union. Tokyo agreed, but it capped its military spending around 1 percent of Japan’s GDP. This was intended to reassure Japan’s neighbors, who viscerally remembered the country’s wartime expansionism. However, because Japan didn’t spend heavily on arms, it had more funds to invest elsewhere.

The question of support for semiconductors was decided by lobbying in Washington. One issue on which Silicon Valley and free market economists agreed was taxes. Bob Noyce testified to Congress in favor of cutting the capital gains tax from 49 percent to 28 percent and advocated loosening financial regulation to let pension funds invest in venture capital firms. After these changes, a flood of money rushed into the venture capital firms on Palo Alto’s Sand Hill Road.

Ishihara had identified a way to coerce America. Japan didn’t need to submit to U.S. demands, Ishihara argued, because America relied on Japanese semiconductors. American military strength, he noted, required Japanese chips.

Disagreements between employees were resolved via a tactic Grove called “constructive confrontation.” His go-to management technique, quipped his deputy Craig Barrett, was “grabbing someone and slamming them over the head with a sledgehammer.”

The system of theft and replication never worked well enough to convince Soviet military leaders they had a steady supply of quality chips, so they minimized the use of electronics and computers in military systems.

popular Soviet joke from the 1980s recounted a Kremlin official who declared proudly, “Comrade, we have built the world’s biggest microprocessor!”

A second issue was overreliance on military customers. The U.S., Europe, and Japan had booming consumer markets that drove chip demand. Civilian semiconductor markets helped fund the specialization of the semiconductor supply chain, creating companies with expertise in everything from ultra-pure silicon wafers to the advanced optics in lithography equipment. The Soviet Union barely had a consumer market, so it produced only a fraction of the chips built in the West.

A final challenge was that the Soviets lacked an international supply chain. Working with America’s Cold War allies, Silicon Valley had forged an ultra-efficient globalized division of labor. Japan led the production of memory chips, the U.S. produced more microprocessors, while Japan’s Nikon and Canon and the Netherland’s ASML split the market for lithography equipment. Workers in Southeast Asia conducted much of the final assembly. American, Japanese, and European companies jostled over their position in this division of labor, but they all benefitted from the ability to spread R&D costs over a far larger semiconductor market than the USSR ever had.

one businessman declined to invest after three meetings with Chang, Taiwan’s prime minister called the stingy executive and reminded him, “The government has been very good to you for the last twenty years. You better do something for the government now.” A check for Chang’s chip foundry arrived soon after. The government also provided generous tax benefits for TSMC, ensuring the company had plenty of money to invest. From day one, TSMC wasn’t really a private business: it was a project of the Taiwanese state.

Chang promised never to design chips, only to build them. TSMC didn’t compete with its customers; it succeeded if they did. A decade earlier, Carver Mead had prophesied a Gutenberg moment in chipmaking, but there was one key difference. The old German printer had tried and failed to establish a monopoly over printing. He couldn’t stop his technology from quickly spreading across Europe, benefitting authors and print shops alike.

Now TSMC had competition from multiple foundries in different countries in East Asia. Singapore’s Chartered Semiconductor, Taiwan’s UMC and Vanguard Semiconductor, and South Korea’s Samsung—which entered the foundry business in 2005—were also competing with TSMC to produce chips designed elsewhere. Most of these companies were subsidized by their governments, but this made chip production cheaper, benefitting the mostly American fabless semiconductor designers they served. Fabless firms, meanwhile, were in the early stages of launching a revolutionary new product chock-full of complex chips: the smartphone.

Dutch lithography company. In 1984, Philips, the Dutch electronics firm, had spun out its internal lithography division, creating ASML.

ASML’s second strength, unexpectedly, was its location in the Netherlands. In the 1980s and 1990s, the company was seen as neutral in the trade disputes between Japan and the United States. U.S. firms treated it like a trustworthy alternative to Nikon and Canon.

Today, nearly every major data center uses x86 chips from either Intel or AMD. The cloud can’t function without their processors.

Spending billions for second place was hardly appealing, especially since Intel’s PC business was still highly profitable and its data center business was growing quickly. So Intel never found a way to win a foothold in mobile devices, which today consume nearly a third of chips sold.

A fixation on hitting short-term margin targets began to replace long-term technology leadership. The shift in power from engineers to managers accelerated this process. Otellini, Intel’s CEO from 2005 to 2013, admitted he turned down the contract to build iPhone chips because he worried about the financial implications. A fixation on profit margins seeped deep into the firm—its hiring decisions, its product road maps, and its R&D processes. The company’s leaders were simply more focused on engineering the company’s balance sheet than its transistors. “It had the technology, it had the people,” one former finance executive at Intel reminisced. “It just didn’t want to take the margin hit.”

Innovators dilemma

By the early 2010s, the most advanced microprocessors had a billion transistors on each chip. The software capable of laying out these transistors was provided by three American firms, Cadence, Synopsys, and Mentor, which controlled around three-quarters of the market. It was impossible to design a chip without using at least one of these firms’ software.

Even he admitted, though, that it was becoming harder to make money while owning and operating a fab. The problem was simple: each generation of technological improvement made fabs more expensive. Morris Chang had drawn a similar conclusion several decades earlier, which is why he thought TSMC’s business model was superior. A foundry like TSMC could fabricate chips for many chip designers, wringing out efficiencies from its massive production volumes that other companies would find difficult to replicate.

Today, building an advanced logic fab costs $20 billion, an enormous capital investment that few firms can afford.

Nvidia was founded in 1993 by Chris Malachowsky, Curtis Priem, and Jensen Huang, the latter of whom remains CEO today. Priem had done fundamental work on how to compute graphics while at IBM, then worked at Sun Microsystems alongside Malachowsky. Huang, who was originally from Taiwan but had moved to Kentucky as a child, worked for LSI, a Silicon Valley chipmaker. He became the CEO and the public face of Nvidia, always wearing dark jeans, a black shirt, and a black leather jacket, and possessing a Steve Jobs−like aura suggesting that he’d seen far into the future of computing.

In 2006, realizing that high-speed parallel computations could be used for purposes besides computer graphics, Nvidia released CUDA, software that lets GPUs be programmed in a standard programming language, without any reference to graphics at all. Even as Nvidia was churning out top-notch graphics chips, Huang spent lavishly on this software effort, at least $10 billion, according to a company estimate in 2017, to let any programmer—not just graphics experts—work with Nvidia’s chips. Huang gave away CUDA for free, but the software only works with Nvidia’s chips. By making the chips useful beyond the graphics industry, Nvidia discovered a vast new market for parallel processing, from computational chemistry to weather forecasting. At the time, Huang could only dimly perceive the potential growth in what would become the biggest use case for parallel processing: artificial intelligence.

So at the depths of the crisis Chang rehired the workers the former CEO had laid off and doubled down on investment in new capacity and R&D. He announced several multibillion-dollar increases to capital spending in 2009 and 2010 despite the crisis. It was better “to have too much capacity than the other way around,” Chang declared. Anyone who wanted to break into the foundry business would face the full force of competition from TSMC as it raced to capture the booming market for smartphone chips. “We’re just at the start,” Chang declared in 2012, as he launched into his sixth decade atop the semiconductor industry.

The revolutionary new phone had many other chips, too: an Intel memory chip, an audio processor designed by Wolfson, a modem to connect with the cell network produced by Germany’s Infineon, a Bluetooth chip designed by CSR, and a signal amplifier from Skyworks,

So the text etched onto the back of each iPhone—“Designed by Apple in California. Assembled in China”—is highly misleading. The iPhone’s most irreplaceable components are indeed designed in California and assembled in China. But they can only be made in Taiwan.

EUV tools work in part because their software works. ASML uses predictive maintenance algorithms to guess when components need to be replaced before they break, for example. It also uses software for a process called computational lithography to print patterns more exactly.

GlobalFoundries was giving up production of new, cutting-edge nodes. It wouldn’t pursue a 7nm process based on EUV lithography, which had already cost $1.5 billion in development and would have required a comparable amount of additional spending to bring online. TSMC, Intel, and Samsung had financial positions that were strong enough to roll the dice and hope they could make EUV work. GlobalFoundries decided that as a medium-sized foundry, it could never make a 7nm process financially viable. It announced it would stop building ever-smaller transistors, slashed R&D spending by a third, and quickly turned a profit after several years of losses. Building cutting-edge processors was too expensive for everyone except the world’s biggest chipmakers. Even the deep pockets of the Persian Gulf royals who owned GlobalFoundries weren’t deep enough. The number of companies capable of fabricating leading-edge logic chips fell from four to three.

China’s government set out a plan called Made in China 2025, which envisioned reducing China’s imported share of its chip production from 85 percent in 2015 to 30 percent by 2025.

as early as 2014, Beijing had decided to double down on semiconductor subsidies, launching what became known as the “Big Fund” to back a new leap forward in chips. Key “investors” in the fund include China’s Ministry of Finance, the state-owned China Development Bank, and a variety of other government-owned firms, including China Tobacco and investment vehicles of the Beijing, Shanghai, and Wuhan municipal governments. Some analysts hailed this as a new “venture capital” model of state support, but the decision to force China’s state-owned cigarette company to fund integrated circuits was about as far from the operating model of Silicon Valley venture capital as could

Integrated circuits made up 15 percent of South Korea’s exports in 2017; 17 percent of Singapore’s; 19 percent of Malaysia’s; 21 percent of the Philippines’; and 36 percent of Taiwan’s.

Softbank had purchased Arm in 2016 for $40 billion, but it sold a 51 percent stake in the China division—which according to Softbank accounted for a fifth of Arm’s global sales—for only $775 million.